An optimized CRISPR/Cas9-based genome editing platform enhances bacteriocin production in Lactiplantibacillus plantarum

Abstract

Efficient genetic engineering of lactic acid bacteria remains technically challenging due to their thick peptidoglycan cell wall, low transformation efficiency, strain-specific restriction–modification systems, and sensitivity to Cas9-induced double-strand breaks. In this study, we adapted an established CRISPR/Cas9 approach for the targeted disruption of plnD, a key negative regulatory gene within the plantaricin quorum-sensing network of Lactiplantibacillus plantarum 8P-A3 through extensive optimization of transformation and genome-editing conditions. The genetically modified strain exhibited upregulation of plnA, plnE, and plnF, accompanied by elevated antimicrobial activity. These findings underscore the feasibility of rationally reconfiguring a quorum-sensing-associated regulatory circuit and provide a practical strategy for successful genetic engineering in L. plantarum for elevated bacteriocin production.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13036-026-00739-5.

Introduction

Lactic acid bacteria (LAB) are widely acknowledged for their generally recognized as safe (GRAS) status, beneficial probiotic properties, and intrinsic ability to synthesize antimicrobial peptides known as bacteriocins. Among these, Lactiplantibacillus plantarum is of particular interest because of its broad-spectrum antimicrobial activity and capacity to thrive in diverse ecological niches, including the complex human skin microbiome [1]. Bacteriocins produced by L. plantarum, termed plantaricins, are ribosomally synthesized peptides that exhibit potent inhibitory activity against several pathogens commonly associated with skin and wound infections [2]. Their membrane-disrupting mode of action, coupled with their narrow resistance profile and selective compatibility with human tissues, collectively positions plantaricins as promising candidates for advanced antimicrobial strategies [3–6].

The rapid rise in antimicrobial resistance (AMR) constitutes a major global health crisis that progressively limits the effectiveness of conventional antibiotics and contributes to persistent infections, delayed wound healing, and an increased healthcare burden [7]. This challenge is particularly pronounced in skin and soft tissue infections (SSTIs), where multidrug-resistant (MDR) pathogens frequently form biofilms that confer protection against host immune defenses and can withstand standard antimicrobial treatments. Biofilm-associated infections often require prolonged or repeated antibiotic therapy and are characterized by high relapse rates. Furthermore, conventional antibiotics frequently exhibit limited penetration into biofilms, leading to treatment failures and chronic infections [8, 9]. The widespread and indiscriminate use of antibiotics has accelerated the emergence of MDR pathogens, necessitating the development of alternative therapeutic strategies such as probiotics [10]. Engineering LAB strains to achieve stronger and more reliable bacteriocin outputs represents an attractive route towards next-generation antimicrobial interventions [11].

In L. plantarum, plantaricin biosynthesis is tightly regulated by a three-component quorum sensing (QS) system [12]. This system comprises an autoinducer peptide PlnA, a membrane-localized histidine kinase PlnB, and two response regulators, PlnC and PlnD [13]. PlnA, a 26-amino-acid cationic peptide, functions as both a bacteriocin-like molecule and an extracellular signaling factor [14–16]. Upon reaching a threshold concentration, PlnA activates PlnB, thereby initiating a phosphorylation cascade that modulates the transcription of multiple bacteriocin operons, including plnABCD, plnNC8 and plnEFI [17–20]. PlnC acts as a transcriptional activator of bacteriocin biosynthesis, whereas PlnD functions as a negative regulator that attenuates PlnA-driven induction [21, 22]. This regulatory architecture is illustrated in Fig. 1, which depicts the PlnA–PlnB–PlnC–PlnD signaling cascade and its control over QS-induced bacteriocin gene expression [23, 24].

Fig. 1.

Quorum sensing–regulated plantaricin system in L. plantarum 8P-A3 strain (Figure adapted and modified from Goel, A. et al., 2023 [23] and Tsapieva, A. et al., 2011 [24]). Schematic representation of the plnABCD regulatory network. Extracellular PlnA activates the histidine kinase PlnB, which phosphorylates the response regulators PlnC and PlnD. PlnC functions as a transcriptional activator of bacteriocin operons, whereas PlnD serves as a negative regulator, attenuating PlnA-driven induction. This QS network governs the density-dependent expression of multiple plantaricin loci, including plnABCD and plnEFI. Created in https://BioRender.com

Early efforts to genetically modify LAB relied on classical mutagenesis (UV irradiation, chemical mutagens), recombineering, mobile group II intron systems, and counter-selectable marker approaches, which improved strain engineering but often suffered from limited precision, low efficiency, labor-intensive screening, and residual genetic scars [25–27]. Genome editing in LAB remains challenging primarily because of their thick peptidoglycan cell wall, low transformation efficiency, strain-specific restriction–modification (RM) systems, and sensitivity to double-strand breaks (DSBs) [28, 29]. To mitigate these barriers, multiple strategies have been developed, including the use of unmethylated plasmid DNA propagated in Dam⁻/Dcm⁻ Escherichia coli hosts, optimized electroporation procedures, in vitro DNA assembly methods, and host-specific transformation workflows [30–33].

The advent of CRISPR/Cas9–assisted recombineering has revolutionized genome engineering in lactobacilli by enabling programmable DNA cleavage and efficient counterselection of unedited cells [34]. However, because lactobacilli possess limited non-homologous end joining (NHEJ) activity, successful editing remains highly dependent on homology-directed repair (HDR) and often exhibits strong strain-to-strain variability [35–42]. Consequently, a wide range of CRISPR-based approaches have been explored, including inducible CRISPR-Cas9 systems, CRISPR interference, RecE/T-assisted editing platforms, alternative Cas variants, modular synthetic biology toolkits, chromosomal knock-in systems, and plasmid-independent ribonucleoprotein delivery strategies [43–58]. Collectively, these studies highlight that no universal genome-editing workflow currently exists for LAB, and successful implementation frequently requires strain-specific optimization of repair templates, expression vectors, plasmid systems, delivery methods, transformation conditions, and host-specific parameters [59–61].

Notably, L. plantarum 8P-A3, an underexplored bacteriocin-rich strain originally investigated for its antimicrobial properties [2, 24], has not previously been subjected to targeted editing of its QS regulatory network. Given the central role of PlnD as a negative response regulator within the plantaricin QS circuit, genetic disruption of plnD represents a rational strategy to relieve negative regulatory control and promote coordinated activation of the plantaricin regulatory network.

This study applies an established two-plasmid CRISPR/Cas9 recombineering workflow to enable targeted disruption of plnD in L. plantarum 8P-A3 [42]. Our approach demonstrates the feasibility of genome engineering in the genetically recalcitrant strain L. plantarum 8P-A3 by extensive optimization and highlights the potential of quorum-sensing regulatory engineering as a strategy to modulate bacteriocin-associated phenotypes in LAB. Expanding the genetic toolbox available for this strain would facilitate mechanistic studies of plantaricin regulation and support future engineering efforts aimed at enhancing antimicrobial performance.

Materials and methods

Bacterial strains, media and growth conditions

L. plantarum 8P-A3 was used as the parental strain for the modification in this study. This strain was kindly provided by Dr. Alexander Suvorov (Institute of Experimental Medicine, Department of Molecular Microbiology, St. Petersburg, Russia) in 2017 and was originally isolated from commercial probiotic preparation Lactobacterinum siccum (Microgen, Moscow, Russia), as previously described [2, 24]. The strain was cultured in De Man, Rogosa and Sharpe (MRS) medium, pH 5.7 (Carl Roth GmbH, Germany, Art. no. HP64.1). When needed, it was grown in MRS medium supplemented with appropriate antibiotics at 37 °C and 150 rpm shaking for 12–16 h under anaerobic conditions. Anaerobic conditions were achieved using a BD GasPak™ EZ gas-generating pouch system with an indicator (Cat no. 260683, BD Company, Sparks MD, USA). Solid media for lactobacilli were prepared using MRS Agar (Carl Roth GmbH, Germany, Art. no. HP65.1). Two solutions, namely MRS-G (MRS-glycine) and saccharose-glycerol (SacGly) solution were prepared for the preparation of electrocompetent L. plantarum cells. The 2.5% MRS-G solution was prepared by adding 0.625 g of glycine to 25 mL of MRS broth, followed by filter sterilization with 0.22 μm filter. The SacGly solution was prepared by dissolving 85.55 g sucrose in 400 mL water, followed by the addition of 100 mL of 50% glycerol. MMRS (modified-MRS) medium was prepared by adding 500 mM sucrose, 20 mM MgCl₂, and 2 mM CaCl₂ to the MRS broth. This was used for cell recovery after electroporation. A MicroPulser Electroporator (Catalog#1652100, Bio-Rad) and Gene Pulser/MicroPulser Electroporation Cuvettes with 0.1 cm gap (Catalog#1652093, Bio-Rad) were used for electroporation. For cloning experiments, NEB 5-alpha competent E. coli cells were used (New England Biolabs GmbH [NEB], Germany, Cat. no. C2987H). This strain was grown in Luria-Bertani (LB) medium (Carl Roth GmbH, Art. no. X968.1) at 37 °C, 250 rpm shaking conditions for 12–16 h with aeration. Solid media for E. coli were prepared using LB Agar (Carl Roth GmbH, Art. no. X969.1). A methyltransferase deficient E. coli strain without a methylation pattern (dam^–/dcm^– Competent E. coli cells) was also received from NEB, Cat. no. C2925H. Super-optimal broth with catabolite repression (SOC) broth/outgrowth medium (NEB, Cat. no. B9020S) was used for cell recovery after transformation, before plating on agar media.

Molecular reagents, enzymes and kits

High-fidelity PCR amplification was performed using Q5^® Hot Start High-Fidelity 2X Master Mix (NEB, Cat no. M0494S), and colony PCR was performed using OneTaq^® Hot Start 2X Master Mix (NEB, Cat no. M0484S). PCR and digestion products were purified using a Monarch^® Spin PCR & DNA Cleanup Kit (NEB, Cat. no. T1130S), and the Monarch^® Spin DNA Gel Extraction Kit (NEB, Cat. no. T1120S). For agarose gel electrophoresis, a 1 kb Plus DNA Ladder (Cat. no. 10787018), and Generuler 100 bp Plus DNA Ladder (Cat. No. SM0321) from ThermoFisher Scientific™, Germany, were used as a reference. DNA restriction enzymes (PvuI, NotI, DpnI, SpeI, HindIII) were purchased from NEB (Cat. nos. R3150S, R3189S, R0176S, R3133S, R3104S, respectively). T4 Polynucleotide Kinase (T4 PNK NEB, Cat. no. M0201S), shrimp alkaline phosphatase (rSAP, NEB Cat. no. M0371S) were used for the phosphorylation and dephosphorylation reactions, respectively. Plasmid DNA was isolated using a QIAprep Spin Miniprep Kit (QIAGEN, Cat. No. 27104) and GeneJET Plasmid Midiprep Kit (ThermoFisher Scientific™, Cat no. K0481). Cell wall lysis for plasmid isolation employed recombinant lysozyme (rlysozyme™), Mutanolysin, and Lysostaphin (Sigma-Aldrich™, Germany, Product nos. 71110-M, M9901, and L9043 respectively). Buffers were prepared using EDTA, NaCl, and Tris·HCl (Sigma-Aldrich™, Cat nos. 60-00-4, 7647-14-5, and 1185-53-1, respectively). Mutagenesis reactions were performed using a Q5^® Site-Directed Mutagenesis Kit (NEB, Cat. no. E0552S). Sticky-end ligation was performed using the Instant Sticky-end Ligase Master Mix (NEB, Cat. no. M0370S). Nuclease-free water (NEB, Cat. no. B1500L) was used for all the reactions. All PCRs were performed using a peqSTAR™ 2X Thermocycler (PEQLAB™), and incubations requiring agitation were performed in a ThermoMixer^® C (Eppendorf™, Germany).

RNA-isolation and reverse-transcription quantitative PCR (RT-qPCR) analysis

Total RNA was extracted using TRIzol™ reagent (Invitrogen™, ThermoFisher Scientific™) combined with mechanical disruption in 0.1-mm zirconia/silica beads (Biospecproducts™, Cat. no. 11079101z; Carl Roth™, Germany) using a Bead Ruptor™ 4, Bead Mill Homogenizer (Omni International). RNA precipitation was performed using ethanol and isopropanol (Cat no. 67-63-0, Merck™ Sigma-Aldrich™), and GlycoBlue™ (Invitrogen™, ThermoFisher Scientific™). Residual genomic DNA was removed using a DNA-free™ DNA Removal Kit (Invitrogen™, ThermoFisher Scientific™). cDNA was synthesized using a SuperScript™ III cDNA synthesis kit (Invitrogen™, ThermoFisher Scientific™) with random hexamers. RT-qPCR reactions were performed in 100 µL MicroAmp™ Fast Optical 96-well reaction plates (AppliedBiosystems™, ThermoFisher Scientific™) using SYBR™ Green dye-KAPA SYBR™ FAST 2X Mastermix (Merck™, Sigma-Aldrich™) and analyzed on a CFX96™ Touch Real-Time PCR Detection System (Bio-Rad), preloaded with relative quantification software (CFX Maestro Software version 2.2). The gene-specific detection primers for plnA, plnE, plnF, and 16S rRNA are listed in Table 1.

Table 1.

List of primers used in this study

2.4. Determination of growth inhibition via monitoring optical density in liquid cultures, and via spot-on-lawn antimicrobial activity assay.

A Tecan Infinite^® M200 PRO microplate reader (Tecan Group Ltd., Männedorf, Switzerland) was used for measuring the optical density at a wavelength of 600 nm (OD₆₀₀) during the growth inhibition assays. Cell-free supernatants (CFS) were collected by centrifugation in a Heraeus Biofuge Primo R refrigerated centrifuge (Thermo Electron Corporation, Waltham, MA, USA). The synthetic mature Plantaricin A peptide (KSSAYSLQMGATAIKQVKKLFKKWGW) was custom-synthesized (Genosphere Biotechnologies) and used as an inducer in the spot-on-lawn assays.

Oligonucleotides and primers

All oligonucleotides and primers were designed using the Geneious Prime^® (www.geneious.com) and Snapgene^® (www.snapgene.com) software packages. These were synthesized by Eurofins Genomics^® (www.eurofinsgenomics.eu) using standard desalting. Site-directed mutagenesis (SDM) primers were purified using HPLC. The complete primer list is presented in Table 1.

Vectors

The following two shuttle vectors were received as gifts from Prof. Dr. Chase L. Beisel (Helmholtz Institute for RNA-based Infection Research, Würzburg, Germany):

1. pCB578 (Addgene # 141095): E. coli–lactobacilli shuttle vector containing SpCas9, tracrRNA, and a repeat-repeat array having an erythromycin (Erm) resistance gene [38, 42].

2. pCB591(Addgene # 141096): E. coli–lactobacilli shuttle vector containing an ampicillin (Amp) resistance gene (E. coli) and a chloramphenicol (Cm) resistance gene (Lactobacillus) [38, 42].

Antibiotics

1. Erythromycin (Duchefa Biochemie, Cat no. 114-07-8): 300 µg/mL in LB medium and 10 µg/mL in MRS medium.

2. Chloramphenicol (Sigma- Aldrich, Cat no. 56-75-7): 34 µg/mL in LB medium and 10 µg/mL in MRS medium.

3. Ampicillin (Duchefa Biochemie, Cat no. 69-52-3): 50 µg/mL in LB medium.

Plasmid construction for expression of crRNA (CR plasmid)

A targeting spacer was designed in Geneious Prime and synthesized as a double-stranded fragment containing a spacer–repeat cassette (20-nt spacer and 36-nt repeat) with flanking overhangs compatible with PvuI and NotI restriction sites. For cloning, the insert was PCR-amplified, and the amplified-insert and pCB578 backbone (Addgene #141095) were digested with PvuI/NotI, and purified separately. To facilitate sticky-end ligation, the vector was dephosphorylated with rSAP, and the insert was phosphorylated with T4 PNK before assembly using an optimized insert: vector ratio (17:1). The ligated products were transformed into E. coli DH5α cells, and recombinant colonies were screened by colony PCR (cPCR) using vector-specific primers (RL323/RL324). Positive clones were confirmed by Sanger sequencing and stored as glycerol stocks. An overview of the workflow for the CR plasmid construction is presented in Fig. 2A.

Fig. 2.

Construction of the CRISPR targeting plasmid (pCB578) and recombineering template plasmid (pCB591). (A) 20-nt targeting spacer (S), flanked by a 36-nt CRISPR repeat (R) and overhangs compatible with PvuI and NotI restriction sites, was cloned into the pCB578 shuttle vector to generate the CRISPR (CR) plasmid expressing the repeat–spacer–repeat array together with SpCas9 and tracrRNA. (B) Separately, the recombineering template (RT) was inserted into the multicloning site (MCS) of the shuttle vector pCB591. Site-directed mutagenesis using the Q5^® mutagenesis kit was then performed on the cloned RT to introduce the programmed deletion/mutation, yielding the final RT plasmid used for homology-directed repair during genome editing. Created in https://BioRender.com

Plasmid construction for recombineering template (RT plasmid)

A plnD-encoding RT containing flanking SpeI and HindIII restriction sites was PCR-amplified using Q5^® polymerase (NEB), and the pCB591 backbone (Addgene #141096) was processed in parallel by digestion with the same restriction enzymes. The amplified template and linearized vector were treated with the DpnI restriction enzyme, purified separately, and then prepared for assembly. The prepared fragments were combined at an optimized insert: vector ratio (13:1) and assembled, and the resulting constructs were transformed into E. coli DH5α. Recombinant plasmids were screened by cPCR (primers: RL402/RL403) and confirmed by Sanger sequencing. Mutations were introduced into the cloned RT using the Q5^® Site-Directed Mutagenesis Kit (NEB), followed by cPCR and sequencing-based verification of the mutant clones. The final sequence-verified RT plasmids were preserved as glycerol stocks. An overview of the RT plasmid construction workflow is shown in Fig. 2B. An overview of the complete CRISPR–Cas9 recombineering workflow used in this study, including RT construction, CR plasmid assembly, and sequential genome editing, is illustrated in Fig. 3.

Fig. 3.

Workflow for CRISPR/Cas9–mediated genome editing in L. plantarum 8P-A3. Schematic overview of the complete genome-editing pipeline adapted for L. plantarum 8P-A3. The workflow is organized into three major stages: (i) construction of the recombineering template (RT) by cloning the plnD gene into pCB591 and introducing the programmed mutation, (ii) assembly of the CRISPR (CR) plasmid pCB578 by inserting the targeting spacer–repeat sequence into the CRISPR array, and (iii) execution of genome editing in L. plantarum through sequential electroporation of the RT and CR plasmids, Cas9-mediated cleavage at the target locus, and homology-directed repair using the RT. Successful edited clones are subsequently isolated after plasmid curing and validated by PCR and sequencing (Figure adapted from Vento et al., 2022 [42]). Created in https://BioRender.com

Preparation of sequence-verified plasmids

The already verified CR and RT plasmids in E. coli DH5α strain were retransformed into a methylation-deficient E. coli dam^–/dcm^– strain to generate unmethylated DNA that was compatible with the L. plantarum strain. Transformants were confirmed using cPCR (CR: RL323/RL324; RT: RL402/RL403) and cultured in LB broth supplemented with erythromycin and ampicillin to obtain CR- and RT-positive clones respectively. The correct plasmids were purified using midiprep columns, quantified, and re-sequenced to confirm spacer integrity (CR) and correct RT mutants before electroporation into L. plantarum.

Preparation of electrocompetent L. plantarum cells

Electrocompetent L. plantarum 8P-A3 cells were prepared by culturing the strain in MRS broth supplemented with 2.5% glycine, followed by multiple washes with ice-cold MgCl₂ and SacGly solution (20% sucrose, 10% glycerol). Cell pellets were resuspended in SacGly solution, aliquoted, flash-frozen in liquid nitrogen, and stored at − 80 °C (detailed protocol provided in the supplementary file).

Electroporation and strain generation

The RT plasmid (5 µL of plasmid DNA ~ 1000 ng µL⁻¹; total DNA amount 5 µg) was transformed into 60 µL of electrocompetent L. plantarum 8P-A3 cells (1.8 kV, 0.1 cm cuvette) by electroporation using an electroporator (MicroPulser Bio-Rad), followed by recovery in MMRS broth and selection on MRS–chloramphenicol (10 µg/mL) agar plate. RT-positive clones were identified by cPCR (RL402/RL403), verified by Sanger sequencing, and used to prepare fresh electrocompetent cells. Similarly, the CR plasmid was subsequently electroporated into these RT plasmid-containing electrocompetent cells, and the transformants were selected on MRS–erythromycin (10 µg/mL) agar plates. Correct CR- and RT-positive clones (colonies carrying both CR and RT plasmids) were confirmed by cPCR targeting the spacer array (RL323/RL324) and Cas9 (JV121/JV123), validated by sequencing, and preserved as glycerol stocks.

Plasmid curing and confirmation of genome editing

The resulting strain was serially passaged for 19 generations in antibiotic-free MRS broth to eliminate both the CR and RT plasmids. The loss of plasmids was verified by replica plating on separate MRS agar plates supplemented with erythromycin (10 µg/mL) and chloramphenicol (10 µg/mL), where the cured colonies displayed complete antibiotic sensitivity. Genomic DNA was extracted (protocol mentioned in the supplementary file), and the entire plantaricin locus was amplified using primers flanking plnA–plnD (FWDGEN/REVGEN – see Table 1). Amplicons were gel-purified and subjected to clonal/linear amplicon NGS sequencing using Oxford Nanopore technology (Eurofins Genomics). The reads were aligned to the reference genome to confirm the presence of the intended chromosomal edits and to assess the locus for any unintended indels or off-target mutations. A stepwise representation of CRISPR/Cas9-based genome editing and mutant selection process for L. plantarum 8P-A3 is shown in Fig. 4.

Fig. 4.

CRISPR–Cas9 genome editing workflow for generating the ΔplnD deletion mutant in L. plantarum 8P-A3. Schematic representation of the stepwise CRISPR–Cas9 recombineering methodology. First, L. plantarum cells are electroporated with the recombineering template (RT) plasmid followed by the CRISPR plasmid carrying SpCas9, tracrRNA, and the targeting spacer. Cas9 introduces a double-strand break at the plnD locus, which is repaired using the RT via homology-directed repair. The colonies appearing on selective agar plate are randomly picked and screened by colony PCR using flanking primers. Edited colonies are identified based on the expected reduced amplicon size, whereas unedited colonies retain the full-length product. Positive clones are subsequently confirmed by sequencing. After validation, both shuttle vectors are cured to obtain a stable ΔplnD mutant strain exhibiting enhanced plnA, plnE, and plnF expression. Created in https://BioRender.com

Calculation of transformation and genome-editing efficiencies

Transformation efficiency was calculated as the number of antibiotic-resistant transformants (colony forming units) obtained per microgram of plasmid DNA electroporated (CFU µg⁻¹ DNA). Genome-editing efficiency was calculated as the percentage of randomly selected transformants that were confirmed to carry the desired mutation in plnD by sequencing following cPCR screening.

Quantification of plnA, plnE, and plnF expression by RT-qPCR

Wild-type (WT) and CRISPR-modified (CRM) strains were grown under anaerobic conditions, and the cells were harvested at 3, 6, 9, and 24 h. RNA was isolated using TRIzol with bead-beating lysis, treated twice with DNase I, and reverse-transcribed using the SuperScript™ III cDNA synthesis kit and random hexamers (the detailed protocol is described in the supplementary file). RT-qPCR was performed using SYBR-based detection on a CFX96™ real-time PCR system (Bio-Rad).

The expression of plnA, plnE, plnF were normalized to that of the 16S rRNA, and the relative expression changes were calculated using the comparative C_t (also known as double delta C_t or 2^−ΔΔCt) method. Three biological replicates with triplicate technical reactions were analyzed, and significance was tested using an unpaired Student’s t-test. A p-value of < 0.05 was considered to be statistically significant.

Photometric growth inhibition assay

Cell-free supernatants (CFS) were collected from 9-h cultures of WT and CRM strains, adjusted to 2 × 10⁶ CFU/mL, and sterile-filtered (0.22 μm). The indicator strain (L. plantarum DSM 16365) was grown to the mid-log phase and adjusted to 2 × 10⁶ CFU/mL. CFS from WT and CRM strains was mixed with indicator culture at a 1:3 ratio in a 96-well microtiter plate, and growth was monitored by OD₆₀₀ every 30 min for 23 h at 37 °C. Growth inhibition was quantified relative to the untreated control. The 1:3 (v/v) ratio of CFS and indicator culture was based on preliminary optimization to maintain measurable indicator growth while allowing detection of inhibitory activity. This assay was used for relative comparison between WT and CRM supernatants under identical conditions rather than for absolute bacteriocin quantification.

Spot-on-lawn assay

Indicator lawns were prepared by mixing an overnight culture of L. plantarum DSM 16365 with molten MRS soft agar (1.2% w/v) and pouring it onto MRS agar plates. Overnight cultures of the WT and CRM strains were pre-mixed with synthetic mature PlnA peptide (26 aa; 250 µM) and 10 µL aliquots were spotted onto the indicator lawn. Plates were incubated anaerobically at 37 °C for 16–18 h, and inhibition zones were examined and measured to assess antimicrobial activity.

Statistical analysis

All experiments were performed with at least three independent biological replicates unless otherwise stated. Where applicable, measurements were additionally performed in technical replicates. Data are presented as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism [version 11.0.0] (GraphPad Software, Boston, MA, USA). Comparisons between two groups were performed using an unpaired two-tailed Student’s t-test. Differences were considered statistically significant at P < 0.05.

Results

Construction and validation of the CRISPR–Cas9 editing system

A dual-plasmid SpCas9 system was designed to enable the targeted disruption of the plnD gene in L. plantarum 8P-A3. The CR plasmid (pCB578) expressed SpCas9, tracrRNA, and a repeat–spacer–repeat cassette, whereas the RT plasmid (pCB591) carried an internal deletion/mutation in the RT.

The correct assembly of both plasmids was confirmed by cPCR, restriction digestion, and Sanger sequencing (Supplementary Figure S1–S4). These validated constructs were subsequently introduced sequentially into L. plantarum 8P-A3 to generate the CRM strain.

Generation and confirmation of the plnD mutant

CRISPR–Cas9 editing using the dual-plasmid system successfully generated a modified L. plantarum 8P-A3 strain carrying a programmed internal deletion/mutation in plnD gene. After serial passaging in antibiotic-free medium, colonies that had lost erythromycin and chloramphenicol resistance were screened and isolated; their antibiotic sensitivity indicated a successful removal of both plasmids. PCR amplification of the full plnA–plnD locus using flanking primers yielded a visibly shorter amplicon of reduced size in the CRISPR-modified (CRM) strain as compared to the wild-type (WT) strain, consistent with an internal deletion within the plnD gene (Fig. 5B).

Fig. 5.

Targeted deletion of plnD using CRISPR–Cas9 and confirmation of genome editing in L. plantarum 8P-A3. (A) Schematic representation of CRISPR–Cas9–mediated genome editing at the plnD locus within the plnABCD operon. The CRISPR plasmid expresses SpCas9 and a spacer targeting the plnD sequence adjacent to the PAM site. Cas9 cleavage induces a double-strand break that is repaired by homologous recombination using the recombineering template (RT), resulting in the programmed deletion/mutation within plnD. The plnD gene sequence is shown in blue, while the edited sequence is shown in red. In the sequence alignment, the ‘-’ region in the CRM sequence indicates the deleted segment, and the bases in the WT sequence highlighted by shading represent deleted or mutated flanking bases. (B) PCR-based screening of wild-type (WT) and CRISPR-modified (CRM) isolates using primers flanking the plnD locus. The CRM strain shows a reduced amplicon size consistent with successful deletion, whereas the WT strain displays the expected full-length product. M: 1 kb DNA ladder. Created in https://BioRender.com

Targeted amplicon sequencing provided nucleotide-level confirmation of the editing. High-stringency alignments were generated independently using validated multi-platform alignment analyses such as Geneious Prime, Benchling, SnapGene and all showed an identical HDR-mediated deletion junction along with some mutated bases within the plnD gene region (Fig. 5A). Importantly, the CRM allele lacked both the complete 20-nt spacer recognition sequence and the adjacent 5′-NGG PAM motif, demonstrating that the Cas9 target site was entirely removed, an accepted hallmark of successful editing in LAB and the mechanistic basis for subsequent plasmid curing.

In addition to the designed deletion, the edited locus contained several substitutions and minor micro-indels clustered near the repair junction. These changes introduced frameshift mutations, ensuring an irreversible loss of PlnD function. Such micro-indels are commonly observed in CRISPR-edited lactobacilli owing to endogenous repair processes. These did not affect adjacent upstream genes (plnA, plnB, and plnC), that remained fully conserved and retained complete identity to the WT locus. Collectively, these analyses confirmed that the CRM strain carries (i) the intended HDR-mediated internal deletion, (ii) complete removal of the CRISPR target site and PAM, and (iii) additional frameshift-generating micro-indels that completely abolish PlnD function. These genomic alterations provide a mechanistic basis for the strong plnA upregulation and enhanced antimicrobial activity observed in subsequent assays.

Quantitative evaluation of transformation efficiency and genome-editing efficiency

To quantitatively assess the performance metrics of the CRISPR/Cas9 editing workflow, transformation and genome-editing efficiencies were determined. Transformation efficiency was expressed as CFU per microgram of plasmid DNA electroporated. To determine genome-editing efficiency, few transformants were randomly selected for downstream analysis. Candidate colonies were initially screened by cPCR, and the presence of the intended edit in plnD was subsequently confirmed by sequencing. The calculated values have been shown in Table 2.

Table 2.

Calculation of transformation efficiency and genome-editing efficiency

Parameter	Value
DNA electroporated (µg)	5
Total transformants obtained	34
Transformation efficiency (CFU µg⁻¹ DNA)	6.8
Number of colonies screened	23
Sequencing-confirmed ΔplnD mutants	7
Genome editing efficiency (%)	30.43

plnD inactivation enhances the expression of bacteriocin-associated genes

To determine whether plnD disruption affects bacteriocin gene transcription, we quantified plnA, plnE, and plnF transcript levels in the WT and CRM strains by using RT-qPCR at different growth phases (3, 6, 9, and 24 h).

Compared to the WT strain, the CRM strain exhibited significantly higher expression of plnA, plnE, and plnF at multiple growth phases (Fig. 6). Among the three genes, plnA showed the strongest induction, followed by plnF and plnE. The mean fold changes (CRM vs. WT) are summarized in Table 3.

Fig. 6.

Relative plnA, plnE, and plnF expression in wild-type and edited strains at different time points. Relative expression of plnA, plnE, and plnF in wild-type (WT) and CRISPR/Cas9-modified (CRM) L. plantarum 8P-A3 strains at different time points. RT-qPCR analysis showing the relative expression of (A) plnA, (B) plnE, and (C) plnF in the CRM strain compared with the WT strain at 3 h, 6 h, 9 h, and 24 h. Relative gene expression was normalized to 16S rRNA reference gene using the 2^−ΔΔCt method. Error bars represent mean ± standard deviation (SD) from three independent biological replicates. Statistical significance was assessed using an unpaired two-tailed Student’s t-test and is indicated by asterisks (* p < 0.05, ** p < 0.01, *** p < 0.001)

Table 3.

Relative expression (2^−ΔΔCt) of plnA, plnE, and plnF in CRM strain relative to the WT strain at different growth phases. Gene expression values are represented as mean ± SD (n = 3 biological replicates)

Time	plnA expression	plnE expression	plnF expression
3 h	27.57 ± 8.01-fold	6.06 ± 3.19-fold	7.96 ± 5.39-fold
6 h	25.33 ± 5.24-fold	6.22 ± 3.36-fold	11.52 ± 3.81-fold
9 h	62.95 ± 17.03-fold	16.85 ± 4.93-fold	27.82 ± 4.63-fold
24 h	7.72 ± 4.02-fold	3.63 ± 1.31-fold	2.99 ± 1.51-fold

Overall, plnA, plnE, and plnF exhibited markedly increased expression in the CRM strain compared with the WT strain. The strongest induction was observed at 9 h for all three genes, suggesting that plnD disruption relieved the negative regulation and enhances the transcription of multiple genes associated with plantaricin production during the mid-exponential to early stationary phases.

plnD inactivation increases antimicrobial activity

Next, we examined whether the elevated expression of plantaricin-associated genes translated into enhanced antimicrobial activity of the CRM strain against the indicator strain L. plantarum DSM 16365.

In the photometric growth inhibition assay, L. plantarum DSM 16,365 culture treated with CRM cell-free supernatant (CFS) showed a significantly greater reduction in indicator growth than those treated with WT CFS (Fig. 7A). The CRM CFS consistently reduced the growth rate and the final optical density of the indicator strain throughout the monitored time course, whereas the WT CFS produced only a modest growth-inhibitory effect. Representative growth curves from three independent biological experiments, all of which exhibited comparable trends. This indicates a substantially higher antimicrobial activity of the CRM strain.

Fig. 7.

Assessment of antimicrobial activity of CRM and WT L. plantarum strains against L. plantarum DSM 16365. (A) Photometric growth inhibition assay: Quantitative measurement of the inhibitory effect of WT and CRM culture supernatants on the growth of L. plantarum DSM 16365 strain. Bacterial growth was monitored by OD₆₀₀ over 23 h. (B) Spot-on-lawn assay: Representative image showing inhibition zones produced by the WT and CRM strains on an agar lawn of the indicator strain. The CRM strain exhibited enhanced antimicrobial activity, as evidenced by a larger clear zone around the spotted colony. Image shown is representative of three independent biological replicates with similar results

Consistent with these quantitative data, spot-on-lawn assays revealed larger and clearer zones of inhibition around CRM spots compared to WT (Fig. 7B). The mean inhibition zone diameters were 11.9 ± 0.7 mm for CRM and 10.3 ± 0.9 mm for WT (n = 3; p < 0.01). No inhibition was observed in the MRS control group.

Together, these results demonstrated that CRISPR/Cas9-mediated inactivation of plnD leads to significantly enhanced bacteriocin-mediated antimicrobial activity in L. plantarum 8P-A3.

Discussion

In this study, we successfully applied an established CRISPR/Cas9-based genome-editing workflow to L. plantarum 8P-A3 and generated a plnD-disrupted mutant strain exhibiting enhanced bacteriocin-associated antimicrobial activity. Targeted inactivation of the QS response regulator plnD resulted in coordinated upregulation of the plantaricin-associated genes plnA, plnE, and plnF, accompanied by enhanced antimicrobial activity. These findings underscore the central regulatory role of the PlnA–PlnB–PlnC–PlnD quorum-sensing network in controlling plantaricin gene expression. Furthermore, our results demonstrate the feasibility of adapting CRISPR/Cas9-mediated genome editing to L. plantarum 8P-A3 and highlight the importance of strain-specific optimization of transformation and editing parameters in genetically recalcitrant LAB strains.

Several critical factors contributed to successful genome editing, including the use of high plasmid dosages, optimized electroporation conditions, and the utilization of Dam⁻/Dcm⁻ E. coli host for plasmid propagation. Collectively, these factors underscore the necessity of strain-specific optimization when implementing CRISPR systems in LAB.

Furthermore, our findings emphasize the critical role of meticulously designed editing reagents. High-purity HPLC-purified primers have been shown to enhance the fidelity of site-directed mutagenesis, whereas strategic spacer design (specifically avoiding potential R-loop structures), has been instrumental in maximizing Cas9 targeting specificity [50, 51]. In LAB, the inability to repair lethal DSBs via NHEJ necessitates reliance on HDR for cell survival [62]. This underscores the need for precise RT design and high-fidelity reagents to ensure recovery of edited cells, aligning with previous CRISPR applications in L. plantarum [38, 42].

At the molecular level, amplicon sequencing revealed that the CRM strain not only harbored the intended deletion in the plnD gene but also exhibited several mutations near the editing site, likely indicative of intrinsic repair dynamics in LAB [23]. The additional sequence variations observed in the vicinity of the edited locus may reflect repair-associated sequence alterations arising during homologous recombination and subsequent selection of edited recombinants. Notably, the CRISPR target site and PAM motif were completely excised, thereby preventing recurrent Cas9 cleavage and facilitating plasmid curing. Because CRISPR/Cas9-mediated editing in lactobacilli relies predominantly on homology-directed repair, such local sequence variations may arise from imperfect recombination events, template switching during repair-template incorporation, incomplete repair-template integration, or selection of edited subpopulations carrying additional target-region modifications. Similar challenges associated with strain-dependent repair outcomes and the requirement for extensive optimization have been reported in CRISPR-mediated genome editing of lactobacilli. Therefore, careful locus-specific verification remains essential when applying genome-editing approaches to genetically recalcitrant LAB strains [35, 38, 41, 47, 48, 63–66]. These confirmed genetic alterations collectively abolished PlnD function, providing a compelling mechanistic explanation for significant upregulation of plnA, together with the downstream bacteriocin genes plnE and plnF, observed using RT-qPCR analysis. This indicates that the disruption of the plnD response regulator activates multiple components of the plantaricin regulatory network. Therefore, the enhanced antimicrobial activity observed in the CRM strain is not solely attributable to increased plnA expression but also to coordinated activation of downstream bacteriocin genes. A mechanistic model summarizing this regulatory rewiring is shown in Fig. 8.

Fig. 8.

Proposed mechanistic model for enhanced bacteriocin production in the CRM strain. Illustration of the regulatory effect of the plnD mutation on expression of plantaricin genes and antimicrobial output. Left: In the WT strain, an intact plnD gene encodes a functional response regulator that represses plnA expression, resulting in basal-level production of plnA, plnE, plnF, thereby limiting bacteriocin secretion. Right: In the CRM strain, the targeted plnD sequence carries a deletion and frameshift-inducing nucleotide substitutions, disrupting normal translation and abolishing PlnD function. Loss of repression leads to upregulated plnA transcription, resulting in elevated secretion of downstream bacteriocins (plnE and plnF). Increased plantaricin output lead to enhanced antimicrobial activity and enhances killing of pathogenic bacteria in the surrounding environment. This model explains the observed upregulation in plnA, plnE, and plnF transcript levels and enhanced antimicrobial activity in the engineered strain. Created in https://BioRender.com

Functionally, plnD inactivation induced a growth phase-modulated increase in plnA, plnE, and plnF transcription, consistent with the removal of a key inhibitory checkpoint within the plantaricin QS regulatory network. The transcript levels of all three genes reached its highest level during the exponential phase, particularly at 9 h, followed by a subsequent decline at 24 h. This temporal expression pattern is consistent with growth-phase-dependent bacteriocin regulation, QS activation thresholds, and feedback inhibition during the stationary phase. The corresponding increase in antimicrobial activity of the CRM strain further supports the functional consequence of this transcriptional activation. It also highlights the potential of regulatory-gene editing as a strategy to reconfigure metabolite production in LAB.

Previous studies have proposed that PlnA activates multiple bacteriocin operons through the PlnB–PlnC/PlnD regulatory cascade. In L. plantarum C11, elevated PlnA signaling induces several bacteriocin operons, including plnEFI, plnJKLR, plnMNOP, and plnGHSTUV, thereby promoting coordinated bacteriocin production. Among these, the two-peptide bacteriocin plantaricin EF act synergistically to form pores in the target cell membrane, resulting in efficient killing of susceptible bacteria [14, 16, 67–72]. Consistent with this hypothesis, our RT-qPCR analysis demonstrated upregulation of both plnE and plnF in CRM strain, indicating activation of the native plnEF operon. These findings provide a plausible mechanistic evidence that plnD inactivation extends beyond derepression of plnA to downstream bacteriocin genes, thereby supporting the observed enhancement in plantaricin-associated antimicrobial activity of the CRM strain.

Although plnA, plnE and plnF transcription increased substantially in the CRM strain, the corresponding increase in antimicrobial activity was comparatively modest. This apparent transcript-to-phenotype disconnect may arise from multiple factors. First, elevated transcript abundance does not necessarily translate proportionally into mature extracellular bacteriocin levels because peptide processing, secretion efficiency, and extracellular stability may become limiting. Second, agar diffusion assays provide only an indirect measure of antimicrobial activity and are strongly influenced by diffusion kinetics, resulting in a nonlinear relationship between bacteriocin concentration and inhibition zone diameter. Finally, because the bacteriocin concentration in the CFS was not directly quantified, the photometric inhibition assay should be interpreted as a comparative functional readout rather than an absolute measure of bacteriocin production. Collectively, these factors may explain why substantial transcriptional activation of the plantaricin regulatory network resulted in only a modest increase in the observed antimicrobial phenotype. Consequently, increased plnA, plnE, and plnF expression may not directly correspond to proportional increase in production of mature, biologically active bacteriocins.

More broadly, this study demonstrates that a previously reported two-plasmid CRISPR/Cas9 system can be successfully implemented in L. plantarum 8P-A3 and may serve as a useful framework for future genome engineering efforts in related LAB strains. Such customizable systems are essential for expanding genetic toolkits across Lactobacillaceae, given the wide variation in editing efficiency across LAB strains [38], and no universal toolkit currently exists. The optimized workflow described herein provides practical blueprint regarding plasmid preparation, transformation, colony screening, and mutant verification for future strain engineering endeavors across the industrial, probiotic, and therapeutic domains.

Despite a strong in vitro antimicrobial phenotype, certain limitations persist. We did not evaluate the in vivo efficacy, microbiome compatibility, or safety, all of which are prerequisites for therapeutic deployment. Genetic modifications may influence colonization dynamics or immunogenicity, necessitating further investigations to validate their stability and performance in complex biological environments. Additionally, although plnD inactivation increased bacteriocin output, its broader transcriptional landscape remains unexplored. Multi-omics analyses would be beneficial for elucidating the comprehensive regulatory impact of plnD disruption. Furthermore, antimicrobial activity was evaluated using a sensitive Lactiplantibacillus indicator strain to establish proof-of-concept for the functional consequences of plnD disruption. Although this approach enabled direct comparison of bacteriocin-associated activity between WT and CRM strains, evaluation against a broader panel of foodborne and clinically relevant pathogens, particularly skin and wound-associated organisms such as Staphylococcus aureus would provide a more comprehensive assessment of the translational potential of the engineered strain.

Conclusion

This study illustrates that CRISPR/Cas9 technology can be effectively applied to L. plantarum 8P-A3, enabling chromosomal mutations in a strain previously considered to be genetically recalcitrant. By incorporating several optimized parameters, we applied an established genome-editing framework that is readily adaptable to other LAB strains with similar genetic barriers.

Future research should investigate the stability, safety, and functional efficacy of these genetically engineered strains in complex microbial communities, as well as extend editing to additional regulatory elements. These engineered LAB strains have significant potential as designer probiotics with tailored therapeutic and biopreservative properties. By leveraging bacteriocin pathways through genetic engineering, this study reinforces the potential of LAB as a customizable microbial chassis for food, biotechnological, and therapeutic applications. In conclusion, this study establishes an optimized CRISPR/Cas9 workflow for targeted editing in L. plantarum 8P-A3 and demonstrates the utility of regulatory gene disruption for modulating bacteriocin-associated phenotypes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

E.H., A.H. and R.L. contributed to the design, visualization, and conceptualization of the research, review, and editing; R.A. contributed to the implementation, investigation, methodology, analyses of the results, and writing of the original draft of the manuscript; L.DL. contributed to the conception, supervision, and revision of the project.

Funding

This study was supported by a research project [IGF 22331 N] of the research association Forschungskuratorium Textil e.V., and supported by the DLR Projektträger within the promotion program “Industrielle Gemeinschaftsforschung” (IGF) of the Federal Ministry for Economic Affairs and Climate Action on the basis of a decision by the German federal parliament (‘‘Bundestag’’). Funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – [541302265] within the Priority Programme SPP 2451 “Engineered Living Materials with Adaptive Functions” funded by Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – [521156679].

Data availability

All data supporting the findings of this study are included in the manuscript and Supplementary Information. Additional raw data such as RT-qPCR values, antimicrobial assay measurements, growth assay data, and sequence validation files are available from the corresponding authors upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.